SearchQuery.php 15.2 KB
Newer Older
1 2 3 4 5
<?php

/**
 * @file
 * Definition of Drupal\search\SearchQuery.
6 7
 *
 * Search query extender and helper functions.
8 9 10 11 12 13 14 15
 */

namespace Drupal\search;

use Drupal\Core\Database\Query\SelectExtender;
use Drupal\Core\Database\StatementEmpty;

/**
16
 * Performs a query on the full-text search index for a word or words.
17 18 19 20 21
 *
 * This function is normally only called by each module that supports the
 * indexed search (and thus, implements hook_update_index()).
 *
 * Results are retrieved in two logical passes. However, the two passes are
22 23
 * joined together into a single query, and in the case of most simple queries
 * the second pass is not even used.
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
 *
 * The first pass selects a set of all possible matches, which has the benefit
 * of also providing the exact result set for simple "AND" or "OR" searches.
 *
 * The second portion of the query further refines this set by verifying
 * advanced text conditions (such as negative or phrase matches).
 *
 * The used query object has the tag 'search_$module' and can be further
 * extended with hook_query_alter().
 */
class SearchQuery extends SelectExtender {
  /**
   * The search query that is used for searching.
   *
   * @var string
   */
  protected $searchExpression;

  /**
43
   * The type of search (search module).
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
   *
   * This maps to the value of the type column in search_index, and is equal
   * to the machine-readable name of the module that implements
   * hook_search_info().
   *
   * @var string
   */
  protected $type;

  /**
   * Positive and negative search keys.
   *
   * @var array
   */
  protected $keys = array('positive' => array(), 'negative' => array());

  /**
   * Indicates whether the first pass query requires complex conditions (LIKE).
   *
63
   * @var bool
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
   */
  protected $simple = TRUE;

  /**
   * Conditions that are used for exact searches.
   *
   * This is always used for the second pass query but not for the first pass,
   * unless $this->simple is FALSE.
   *
   * @var DatabaseCondition
   */
  protected $conditions;

  /**
   * Indicates how many matches for a search query are necessary.
   *
   * @var int
   */
  protected $matches = 0;

  /**
   * Array of search words.
   *
   * These words have to match against {search_index}.word.
   *
   * @var array
   */
  protected $words = array();

  /**
   * Multiplier for the normalized search score.
   *
   * This value is calculated by the first pass query and multiplied with the
   * actual score of a specific word to make sure that the resulting calculated
   * score is between 0 and 1.
   *
   * @var float
   */
  protected $normalize;

  /**
   * Indicates whether the first pass query has been executed.
   *
107
   * @var bool
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
   */
  protected $executedFirstPass = FALSE;

  /**
   * Stores score expressions.
   *
   * @var array
   */
  protected $scores = array();

  /**
   * Stores arguments for score expressions.
   *
   * @var array
   */
  protected $scoresArguments = array();

  /**
   * Total value of all the multipliers.
   *
   * @var array
   */
  protected $multiply = array();

  /**
   * Whether or not search expressions were ignored.
   *
   * The maximum number of AND/OR combinations exceeded can be configured to
   * avoid Denial-of-Service attacks. Expressions beyond the limit are ignored.
   *
138
   * @var bool
139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202
   */
  protected $expressionsIgnored = FALSE;

  /**
   * Sets up the search query expression.
   *
   * @param $query
   *   A search query string, which can contain options.
   * @param $module
   *   The search module. This maps to {search_index}.type in the database.
   *
   * @return
   *   The SearchQuery object.
   */
  public function searchExpression($expression, $module) {
    $this->searchExpression = $expression;
    $this->type = $module;

    return $this;
  }

  /**
   * Applies a search option and removes it from the search query string.
   *
   * These options are in the form option:value,value2,value3.
   *
   * @param $option
   *   Name of the option.
   * @param $column
   *   Name of the database column to which the value should be applied.
   *
   * @return
   *   TRUE if a value for that option was found, FALSE if not.
   */
  public function setOption($option, $column) {
    if ($values = search_expression_extract($this->searchExpression, $option)) {
      $or = db_or();
      foreach (explode(',', $values) as $value) {
        $or->condition($column, $value);
      }
      $this->condition($or);
      $this->searchExpression = search_expression_insert($this->searchExpression, $option);
      return TRUE;
    }
    return FALSE;
  }

  /**
   * Parses the search query into SQL conditions.
   *
   * We build two queries that match the dataset bodies.
   */
  protected function parseSearchExpression() {
    // Matchs words optionally prefixed by a dash. A word in this case is
    // something between two spaces, optionally quoted.
    preg_match_all('/ (-?)("[^"]+"|[^" ]+)/i', ' ' .  $this->searchExpression , $keywords, PREG_SET_ORDER);

    if (count($keywords) ==  0) {
      return;
    }

    // Classify tokens.
    $or = FALSE;
    $warning = '';
203
    $limit_combinations = config('search.settings')->get('and_or_limit');
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
    // The first search expression does not count as AND.
    $and_count = -1;
    $or_count = 0;
    foreach ($keywords as $match) {
      if ($or_count && $and_count + $or_count >= $limit_combinations) {
        // Ignore all further search expressions to prevent Denial-of-Service
        // attacks using a high number of AND/OR combinations.
        $this->expressionsIgnored = TRUE;
        break;
      }
      $phrase = FALSE;
      // Strip off phrase quotes.
      if ($match[2]{0} == '"') {
        $match[2] = substr($match[2], 1, -1);
        $phrase = TRUE;
        $this->simple = FALSE;
      }
      // Simplify keyword according to indexing rules and external
      // preprocessors. Use same process as during search indexing, so it
      // will match search index.
      $words = search_simplify($match[2]);
      // Re-explode in case simplification added more words, except when
      // matching a phrase.
      $words = $phrase ? array($words) : preg_split('/ /', $words, -1, PREG_SPLIT_NO_EMPTY);
      // Negative matches.
      if ($match[1] == '-') {
        $this->keys['negative'] = array_merge($this->keys['negative'], $words);
      }
      // OR operator: instead of a single keyword, we store an array of all
      // OR'd keywords.
      elseif ($match[2] == 'OR' && count($this->keys['positive'])) {
        $last = array_pop($this->keys['positive']);
        // Starting a new OR?
        if (!is_array($last)) {
          $last = array($last);
        }
        $this->keys['positive'][] = $last;
        $or = TRUE;
        $or_count++;
        continue;
      }
      // AND operator: implied, so just ignore it.
      elseif ($match[2] == 'AND' || $match[2] == 'and') {
        $warning = $match[2];
        continue;
      }

      // Plain keyword.
      else {
        if ($match[2] == 'or') {
          $warning = $match[2];
        }
        if ($or) {
          // Add to last element (which is an array).
          $this->keys['positive'][count($this->keys['positive']) - 1] = array_merge($this->keys['positive'][count($this->keys['positive']) - 1], $words);
        }
        else {
          $this->keys['positive'] = array_merge($this->keys['positive'], $words);
          $and_count++;
        }
      }
      $or = FALSE;
    }

    // Convert keywords into SQL statements.
    $this->conditions = db_and();
    $simple_and = FALSE;
    $simple_or = FALSE;
    // Positive matches.
    foreach ($this->keys['positive'] as $key) {
      // Group of ORed terms.
      if (is_array($key) && count($key)) {
        $simple_or = TRUE;
        $any = FALSE;
        $queryor = db_or();
        foreach ($key as $or) {
          list($num_new_scores) = $this->parseWord($or);
          $any |= $num_new_scores;
          $queryor->condition('d.data', "% $or %", 'LIKE');
        }
        if (count($queryor)) {
          $this->conditions->condition($queryor);
          // A group of OR keywords only needs to match once.
          $this->matches += ($any > 0);
        }
      }
      // Single ANDed term.
      else {
        $simple_and = TRUE;
        list($num_new_scores, $num_valid_words) = $this->parseWord($key);
        $this->conditions->condition('d.data', "% $key %", 'LIKE');
        if (!$num_valid_words) {
          $this->simple = FALSE;
        }
        // Each AND keyword needs to match at least once.
        $this->matches += $num_new_scores;
      }
    }
    if ($simple_and && $simple_or) {
      $this->simple = FALSE;
    }
    // Negative matches.
    foreach ($this->keys['negative'] as $key) {
      $this->conditions->condition('d.data', "% $key %", 'NOT LIKE');
      $this->simple = FALSE;
    }

    if ($warning == 'or') {
      drupal_set_message(t('Search for either of the two terms with uppercase <strong>OR</strong>. For example, <strong>cats OR dogs</strong>.'));
    }
  }

  /**
   * Helper function for parseQuery().
   */
  protected function parseWord($word) {
    $num_new_scores = 0;
    $num_valid_words = 0;
    // Determine the scorewords of this word/phrase.
    $split = explode(' ', $word);
    foreach ($split as $s) {
      $num = is_numeric($s);
326
      if ($num || drupal_strlen($s) >= config('search.settings')->get('index.minimum_word_size')) {
327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
        if (!isset($this->words[$s])) {
          $this->words[$s] = $s;
          $num_new_scores++;
        }
        $num_valid_words++;
      }
    }
    // Return matching snippet and number of added words.
    return array($num_new_scores, $num_valid_words);
  }

  /**
   * Executes the first pass query.
   *
   * This can either be done explicitly, so that additional scores and
   * conditions can be applied to the second pass query, or implicitly by
   * addScore() or execute().
   *
   * @return
   *   TRUE if search items exist, FALSE if not.
   */
  public function executeFirstPass() {
    $this->parseSearchExpression();

    if (count($this->words) == 0) {
352
      form_set_error('keys', format_plural(config('search.settings')->get('index.minimum_word_size'), 'You must include at least one positive keyword with 1 character or more.', 'You must include at least one positive keyword with @count characters or more.'));
353 354 355
      return FALSE;
    }
    if ($this->expressionsIgnored) {
356
      drupal_set_message(t('Your search used too many AND/OR expressions. Only the first @count terms were included in this search.', array('@count' => config('search.settings')->get('and_or_limit'))), 'warning');
357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
    }
    $this->executedFirstPass = TRUE;

    if (!empty($this->words)) {
      $or = db_or();
      foreach ($this->words as $word) {
        $or->condition('i.word', $word);
      }
      $this->condition($or);
    }
    // Build query for keyword normalization.
    $this->join('search_total', 't', 'i.word = t.word');
    $this
      ->condition('i.type', $this->type)
      ->groupBy('i.type')
      ->groupBy('i.sid')
      ->having('COUNT(*) >= :matches', array(':matches' => $this->matches));

    // Clone the query object to do the firstPass query;
    $first = clone $this->query;

    // For complex search queries, add the LIKE conditions to the first pass query.
    if (!$this->simple) {
      $first->join('search_dataset', 'd', 'i.sid = d.sid AND i.type = d.type');
      $first->condition($this->conditions);
    }

    // Calculate maximum keyword relevance, to normalize it.
    $first->addExpression('SUM(i.score * t.count)', 'calculated_score');
    $this->normalize = $first
      ->range(0, 1)
      ->orderBy('calculated_score', 'DESC')
      ->execute()
      ->fetchField();

    if ($this->normalize) {
      return TRUE;
    }
    return FALSE;
  }

  /**
   * Adds a custom score expression to the search query.
   *
   * Each score expression can optionally use a multiplier, and multiple
   * expressions are combined.
   *
   * @param $score
   *   The score expression.
   * @param $arguments
   *   Custom query arguments for that expression.
   * @param $multiply
   *   If set, the score is multiplied with that value. Search query ensures
   *   that the search scores are still normalized.
411 412 413
   *
   * @return object
   *   The updated query object.
414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518
   */
  public function addScore($score, $arguments = array(), $multiply = FALSE) {
    if ($multiply) {
      $i = count($this->multiply);
      $score = "CAST(:multiply_$i AS DECIMAL) * COALESCE(( " . $score . "), 0) / CAST(:total_$i AS DECIMAL)";
      $arguments[':multiply_' . $i] = $multiply;
      $this->multiply[] = $multiply;
    }

    $this->scores[] = $score;
    $this->scoresArguments += $arguments;

    return $this;
  }

  /**
   * Executes the search.
   *
   * If not already done, this executes the first pass query. Then the complex
   * conditions are applied to the query including score expressions and
   * ordering.
   *
   * @return
   *   FALSE if the first pass query returned no results, and a database result
   *   set if there were results.
   */
  public function execute()
  {
    if (!$this->executedFirstPass) {
      $this->executeFirstPass();
    }
    if (!$this->normalize) {
      return new StatementEmpty();
    }

    // Add conditions to query.
    $this->join('search_dataset', 'd', 'i.sid = d.sid AND i.type = d.type');
    $this->condition($this->conditions);

    if (empty($this->scores)) {
      // Add default score.
      $this->addScore('i.relevance');
    }

    if (count($this->multiply)) {
      // Add the total multiplicator as many times as requested to maintain
      // normalization as far as possible.
      $i = 0;
      $sum = array_sum($this->multiply);
      foreach ($this->multiply as $total) {
        $this->scoresArguments[':total_' . $i] = $sum;
        $i++;
      }
    }

    // Replace i.relevance pseudo-field with the actual, normalized value.
    $this->scores = str_replace('i.relevance', '(' . (1.0 / $this->normalize) . ' * i.score * t.count)', $this->scores);
    // Convert scores to an expression.
    $this->addExpression('SUM(' . implode(' + ', $this->scores) . ')', 'calculated_score', $this->scoresArguments);

    if (count($this->getOrderBy()) == 0) {
      // Add default order after adding the expression.
      $this->orderBy('calculated_score', 'DESC');
    }

    // Add tag and useful metadata.
    $this
      ->addTag('search_' . $this->type)
      ->addMetaData('normalize', $this->normalize)
      ->fields('i', array('type', 'sid'));

    return $this->query->execute();
  }

  /**
   * Builds the default count query for SearchQuery.
   *
   * Since SearchQuery always uses GROUP BY, we can default to a subquery. We
   * also add the same conditions as execute() because countQuery() is called
   * first.
   */
  public function countQuery() {
    // Clone the inner query.
    $inner = clone $this->query;

    // Add conditions to query.
    $inner->join('search_dataset', 'd', 'i.sid = d.sid AND i.type = d.type');
    $inner->condition($this->conditions);

    // Remove existing fields and expressions, they are not needed for a count
    // query.
    $fields =& $inner->getFields();
    $fields = array();
    $expressions =& $inner->getExpressions();
    $expressions = array();

    // Add the sid as the only field and count them as a subquery.
    $count = db_select($inner->fields('i', array('sid')), NULL, array('target' => 'slave'));

    // Add the COUNT() expression.
    $count->addExpression('COUNT(*)');

    return $count;
  }
}